--- title: "Heavy spend, muted returns: Has AMZN become NVDA's workhorse?" type: "Topics" locale: "en" url: "https://longbridge.com/en/dolphin/post/44651352.md" description: "Near-term pressure and heavy cash burn. However, valuation now looks undemanding." datetime: "2026-10-30T02:35:57.000Z" locales: - [en](https://longbridge.com/en/dolphin/post/44651352.md) - [zh-CN](https://longbridge.com/zh-CN/dolphin/post/44651352.md) - [zh-HK](https://longbridge.com/zh-HK/dolphin/post/44651352.md) author: "[Dolphin Research](https://longbridge.com/en/dolphin.md)" generator: "portal-rs" --- # Heavy spend, muted returns: Has AMZN become NVDA's workhorse? $Amazon(AMZN.US) released Q3 results early on the 30th Beijing time. Dolphin Research's read is neutral-to-bearish: the three metrics the market cares most about all missed. AWS growth came in below buy-side hopes, where bulls wanted >43% growth and >40% margin, and neither was met. Ex-Prime Day, ads still slowed markedly. Capex far exceeded expectations, driving free cash flow deep into negative territory, and management said 2027 capex will be well above this year. The only bright spot was profit, which again topped the upper end of guidance. **1、Overall: revenue in line, profit well above the top end of guidance** Total revenue was $201.9bn, +12.1% YoY (+13% ex-FX), broadly in line with Bloomberg consensus. The growth rate slowed sharply from 19.6% in Q2, mainly because Prime Day was pulled forward to Q2, which dragged Q3 growth by roughly 4ppt, with FX shaving another ~0.5ppt. Operating profit was $28.1bn, +61% YoY (the base was low due to one-offs such as the FTC settlement last year). It beat consensus by about 8% and again topped prior guidance. OPM was 13.9%, a record high, up 20bps QoQ. **2、AWS: continued acceleration, but short of the ~43% buy-side 'pass line'** AWS revenue was $46.0bn, +39.3% YoY, accelerating by about 250bps vs. Q2. However, buy-side expectations were ~43% (range 42%-44%), implying a miss of ~3-4ppt; sequential net adds were $3.75bn, below Q2's $4.65bn, so the incremental revenue shrank. Management attributed this to uneven capacity go-lives across quarters. The positive is management indicated AWS growth above 40% in 2027 and 2028. That aligns with big-bank expectations for 2027 but is well above the pre-print ~35% for 2028. Segment OP was $17.8bn, ~12% above consensus. Margin was 38.7%, down 70bps QoQ and below the ~40% buy-side threshold, but last quarter's margin was abnormally high due to a ~$0.6bn one-off from energy contracts; ex that it was ~38.1%, so this quarter's margin actually improved QoQ. Call it underwhelming growth with margin support. On the call, backlog at quarter-end was $578.0bn, up $82bn QoQ and triple-digit YoY, with customer concentration described as dispersed. AWS has added roughly $100bn of net new orders for three straight quarters, and backlog now stands at 3.1x AWS annualized revenue, underpinning the mid-to-long term growth outlook. **3、Advertising: clear slowdown even after adding back Prime Day** The largest retail growth driver—ad revenue—was $19.6bn, +10.9% YoY (+12% ex-FX), about 6% below consensus, with a sharp deceleration vs. Q2. Management explained: Prime Day moved to Q2, creating an ~850bps sequential drag; Q2 also had NBA's first-season games, which Q3 did not. While the company guided Q4 ad growth back to the high-teens, that remains below the 20%+ pace of prior quarters. Channel checks suggest two drivers for continued ad slowdown. First, merchants' total cost burden (take rate, fulfillment, ads) is near the ceiling, capping on-site search ad growth and requiring video, Twitch and other non-transaction channels to pull more weight. Second, AI assistants, though likely a limited factor, are also diverting traffic from the marketplace to AI. Elsewhere in retail, online stores revenue was $68.5bn, +1.6% YoY, also slowing and below expectations. Third-party seller services and subscriptions grew more steadily and beat the market. Why didn't subscriptions and 3P services weaken? Subscriptions are recognized monthly and are less affected by Prime Day timing, while 3P recently added fuel and logistics surcharges to support monetization. By raising the take rate, they offset the impact, which also shows merchants' burden is indeed rising. **4、Retail by region: NA margin above expectations, Intl lagged** By region, NA retail revenue was $112.9bn, +6.2% YoY, in line with consensus. Intl retail revenue was $43.1bn, +5.4% YoY (+8% ex-FX), with nominal growth slightly below expectations. Versus the broader industry, U.S. Census (seasonally adjusted) shows nonstore retail up 7.5% YoY in Q3, down from 9.2% in Q2, with Q2+Q3 combined at +8.3%. Amazon's NA segment grew about +11% over Q2+Q3 combined, ~300bps above nonstore retail, indicating continued share gains. On profit, NA OP was $8.72bn with a 7.7% margin, ~110bps above consensus. The margin appears lower QoQ, but last quarter benefited from a one-off tax refund; ex that, this quarter's margin improved QoQ. Intl OP was $1.58bn with a 3.7% margin, below the 4.5% consensus, with FX a ~$0.13bn drag. **5、Capex far above expectations, more in 2027, FCF deeply underwater** Cash capex was $77.3bn, 31% above consensus and up 43% QoQ, a major surprise. Management did not directly specify the drivers, saying spend pacing follows demand visibility. We think price hikes in storage and other hardware, plus payment timing, may be factors. Operating cash flow was $42.4bn, +19% YoY, but FCF was -$33.5bn, deeply negative and far below expectations. With AWS growth not particularly surprising, the clearly worse cash flow raises questions on ROI and concerns about funding pressure on the balance sheet or shareholders. Indeed, Amazon has been financing more frequently, adding about $48.1bn of long-term debt this quarter. By full-year guide, Q4 capex is about $49.0bn, well below the roughly $65.0bn consensus and back to a more normal level, suggesting this quarter was an outlier. Yet management also said 2027 capex will be 'significantly higher' than 2026 (pre-print buy-side expected ~$325bn-$350bn), implying even greater pressure on cash flow and financing ahead. **6、GPM and expenses: GPM beat, tech and infrastructure spend elevated** GPM was 53.2%, about 110bps above consensus, as mix continued to tilt toward higher-margin businesses like AWS and 3P services, driving a structural uptrend in GPM. On costs, tech and infrastructure expense was $34.8bn, ~3% above consensus, the only clear overspend and consistent with broad-based AI R&D investment. Other expenses were generally below expectations. Marketing was flat YoY and G&A declined YoY, similar to Microsoft, indicating aggressive control and even cuts outside AI and other tech areas. **7、Q4 guide: revenue in line, profit upper bound slightly above consensus but below buy-side** Revenue guidance is $240.0-$247.0bn, +12%-16% YoY, with the midpoint ~0.4% below consensus but essentially in line, and with FX expected to be a ~60bps drag. The upper bound only matches the lower end of buy-side expectations (~$247bn-$248bn). Operating profit guidance is $28.5-$33.5bn. Per usual we look at the upper bound, which is slightly above consensus but below buy-side expectations of ~$34bn-$35bn; the implied OPM is about 13.6%, up roughly 190bps YoY. Back-solving implies retail growth is still a bit light: assuming AWS is +40% in Q4, retail revenue would be +7%-11% YoY, with a midpoint of ~+9%, below the ~11%-12% overall pace in the first three quarters. Hitting the upper bound would only keep it roughly flat. **8、Core metrics at a glance; see call transcript**[**link**](https://longbridge.com/zh-CN/dolphin/post/44644977) **Dolphin Research view:** Overall, the signals skew negative, with the three most-watched items all missing: AWS growth failed to clear the bar and sequential net adds shrank; retail's biggest profit driver—ads—slowed materially; capex surged, FCF went deeply negative, and spend will rise again next year. The call did not dispel these concerns, as management did not directly address the slowdown in incremental growth or capex returns. Next-quarter guidance is in line with Bloomberg consensus and not bad, but it fell short of more bullish buy-side hopes. Beyond the print, we focus on the following: **1) AWS: no near-term upside surprise, still high visibility mid-to-long term** Near term, AWS did not exceed expectations: growth missed the buy-side bar and sequential net adds shrank. Management waved this off with 'uneven capacity go-lives', and noted that a large amount of capacity will come online over the next 8-10 quarters, making growth likely track capacity and limiting upside surprises in the short run. Mid-to-long term visibility is high: backlog exceeds 3x AWS annualized revenue, with large chunks of capacity booked beyond 2027. Management reiterated growth above 40% in 2027 and 2028, well above prior big-bank expectations for 2028. The bulk of orders comes from two large contracts with OpenAI and Anthropic (about half), which implies higher concentration but also contracted demand. In-house chips are growing triple digits YoY, and Trainium will be sold externally by end-2027, offering additional upside optionality. We think the issue is not demand but supply timing and overly high market expectations: near-term growth is unlikely to surprise, but with capacity ramping as planned, roughly 40% growth mid-to-long term is supported by orders. **2) Compute pricing power: Bedrock and price hikes follow the same logic** On distribution, Anthropic's IPO materials indicate about half of its H1 revenue flowed through cloud channels, highlighting that cloud vendors' distribution matters even for top AI labs. On pricing, news reports suggest AWS reserved GPU capacity prices have risen for four straight quarters, up a cumulative 50%-70%, indicating growing pricing power. This suggests cloud vendors control both scarce compute and distribution, and their bargaining power across the value chain is rising. Bedrock moves AWS from selling compute to selling inference services, with most inference running on in-house chips and margins above AWS overall, which is why AWS margins have not been dragged by AI even as revenue accelerates. **3) Advertising: monetization near a ceiling; teens may be the new normal** As noted, on-site search ads are constrained by merchants' cost burden. More importantly, digital ads overall outperformed in Q3 while Amazon slowed, implying the issue is retail ad monetization nearing its ceiling rather than industry cycles. Over time, AI shopping assistants (including Amazon's Alexa) help users skip search and display pages, structurally reducing ad inventory. The more successful the assistant, the greater the diversion from search ads. We think ad growth is converging toward GMV growth. A move from 20%+ to the mid-teens could be the new normal. **4) Capex and financing: more next year, returns not quantified** Management said 2027 capex will be 'significantly higher' but did not quantify the scale or returns. In today's environment, investors clearly dislike high capex and cash burn. Amazon has conducted multiple financings this year, mainly via bonds. New long-term borrowings in Q1-Q3 were about $53.4bn, $13.6bn and $48.1bn, respectively, with quarter-end long-term debt around $173.1bn; per Bloomberg, Amazon is among the largest U.S. corporate bond issuers this year. Based on the pre-print buy-side view of 2027 capex at ~$325bn-$350bn and consensus operating cash flow of about $247bn, we roughly estimate a 2027 FCF gap of ~$80bn-$100bn. That equals 70%-90% of quarter-end cash and marketable securities (~$114.2bn); including follow-on commitments for external investments, the gap is larger, making another year of large bond issuance almost inevitable. Net-net, this print suffers from 'not good enough near term, spending too aggressively': AWS and ads did not deliver the upside the buy-side wanted, while capex and financing are accelerating. The crowded long positioning ahead of the print will take time to unwind. That said, AWS fundamentals mid-to-long term have not deteriorated: backlog exceeding 3x annualized revenue and a clear capacity plan underpin growth, and the 40%+ growth guide for 2027 and 2028 is above prior big-bank expectations. Retail remains relatively steady with early signs of softening: the most profitable ads business is slowing, which may cap further retail margin expansion. **Earnings forecast changes:** we update the model with Q3 actuals and Q4 guidance, and, per usual, take slightly above the guidance upper bound for Q4 OP (~$34.5bn). The key change is lifting 2028 AWS growth from ~35% pre-print big-bank average to ~41%, aligning with management's 'above 40% in 2027 and 2028'. Post-update, 2027 OP is about $154.2bn and 2028 about $209.5bn. **Valuation:** we keep a sum-of-the-parts for retail and AWS, roll to 2027 and lower multiples (previously 25x for retail and 40x for AWS). For retail, we apply 20x PE on ~22% OP growth, implying about $940bn; for AWS, 30x on ~44% OP growth implies about $2.4tn (both on after-tax OP, with a 17% tax rate). **Neutral FV:** totals about $3.4tn, or roughly $312/sh. At about $224 after-hours, the current market cap is around $2.4tn, ~28% below our neutral FV. On the same basis, that equates to ~14x for retail and ~22x for AWS on 2027 after-tax OP. Looking further out, the legacy businesses of cloud giants—software/SaaS, search/ads and retail—will all face AI pressure. The common logic is core under pressure, offset by AI and cloud. Among the three, we think retail may face the smallest ultimate hit: fulfillment involves the physical world, and Amazon's logistics network raises the moat further. Ads—the 'discovery' layer—faces the biggest impact, which is why the ad slowdown this quarter warrants ongoing monitoring. Therefore, we think the stock will remain capped near term by the capex and financing narrative. But among the cloud Big 3, Amazon's legacy business is relatively more resilient to AI, and combined with AWS's order visibility, the risk-reward should improve as sentiment resets. \ **Risk disclosures and statements:**[**Dolphin Research disclaimer and general disclosure**](https://support.longbridge.global/topics/misc/dolphin-disclaimer) Past Dolphin Research coverage on Amazon: Earnings takeaways Jul 31, 2026 print recap '[AWS surges: Amazon finally catches up with the pack](https://longbridge.com/en/dolphin/post/43114079)' Jul 31, 2026 call '[Amazon (Trans): CAPEX raised to $220bn](https://longbridge.com/en/dolphin/post/43096142)' Apr 30, 2026 call 'Amazon (Trans): A once-in-a-lifetime AI investment' Apr 30, 2026 print recap 'Retail steady, AI breaks through: is Amazon back in the top tier?' Feb 6, 2026 print recap 'A $200bn arms race: Amazon cranks AI to new heights' Feb 6, 2026 call 'Amazon (Trans): all new capacity was absorbed immediately' Oct 30, 2025 call '[Amazon (Trans): compute supply to double again by 2027](https://longbridge.com/en/topics/35834740)' Oct 30, 2025 print recap 'AWS big reversal: has Amazon finally turned the corner?' Deep dives Dec 18, 2024 '[Amazon e-com endgame: retail wrapper, ad soul?](https://longportapp.cn/en/topics/26148582)' ### Related Stocks - [AMZN.US](https://longbridge.com/en/quote/AMZN.US.md) - [NVDA.US](https://longbridge.com/en/quote/NVDA.US.md) - [AMZD.US](https://longbridge.com/en/quote/AMZD.US.md) - [AMZU.US](https://longbridge.com/en/quote/AMZU.US.md) - [AMZZ.US](https://longbridge.com/en/quote/AMZZ.US.md) - [AMZY.US](https://longbridge.com/en/quote/AMZY.US.md) - [AMZP.US](https://longbridge.com/en/quote/AMZP.US.md) - [AMZO.US](https://longbridge.com/en/quote/AMZO.US.md) - [AMZW.US](https://longbridge.com/en/quote/AMZW.US.md) - [NVDD.US](https://longbridge.com/en/quote/NVDD.US.md) --- > **Disclaimer: This article is for reference only and does not constitute any investment advice.**